Additive Manufacturing of Inconel 718 using Electron Beam Melting: Processing, Post-Processing, & Mechanical Properties

(1)

ADDITIVE MANUFACTURING OF INCONEL 718

USING ELECTRON BEAM MELTING:

PROCESSING, POST-PROCESSING, & MECHANICAL PROPERTIES

A Dissertation by

WILLIAM JAMES SAMES V

Submitted to the Office of Graduate and Professional Studies of Texas A&M University

in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY

Chair of Committee, Sean M. McDeavitt Committee Members, Raymundo Arroyave

Ryan R. Dehoff Delia Perez-Nunez Lin Shao

Head of Department, Yassin Hassan

May 2015

Major Subject: Nuclear Engineering

(2)

ii

ABSTRACT

Additive Manufacturing (AM) process parameters were studied for production of the high temperature alloy Inconel 718 using Electron Beam Melting (EBM) to better understand the relationship between processing, microstructure, and mechanical properties. Processing parameters were analyzed for impact on process time, process temperature, and the amount of applied energy. The applied electron beam energy was shown to be integral to the formation of swelling defects. Standard features in the microstructure were identified, including previously unidentified solidification features such as shrinkage porosity and non-equilibrium phases. The as-solidified structure does not persist in the bulk of EBM parts due to a high process hold temperature (~1000°C), which causes in situ homogenization. The most significant variability in as-fabricated microstructure is the formation of intragranular delta-phase needles, which can form in samples produced with lower process temperatures (< 960°C). A novel approach was developed and demonstrated for controlling the

temperature of cool down, thus providing a technique for in situ heat treatment of material. This technique was used to produce material with hardness of 478±7 HV with no

processing, which exceeds the hardness of peak-aged Inconel 718. Traditional post-processing methods of hot isostatic pressing (HIP) and solution treatment and aging (STA) were found to result in variability in grain growth and phase solution. Recrystallization and grain structure are identified as possible mechanisms to promote grain growth. These results led to the conclusion that the first step in thermal post-processing of EBM Inconel 718 should be an optimized solution treatment to reset phase variation in the as-fabricated microstructure without incurring significant grain growth. Such an optimized solution treatment was developed (1120°C, 2hr) for application prior to aging or HIP. The majority of as-fabricated tensile properties met ASTM AM Inconel 718 standards for yield stress and ultimate tensile strength, and STA yield stress, ultimate tensile strength, and elongation exceeded the ASTM standards for AM Inconel 718.

(3)

iii

ACKNOWLEDGEMENTS

I have received support from many colleagues throughout this work, to which I am very thankful.

I would like to thank my graduate advisor, Sean McDeavitt, for his guidance and for giving me the freedom to perform my research at ORNL, in a unique arrangement. His guidance has been far-sighted and enabled me to think creatively in developing my degree program.

I would like to thank my ORNL mentor, Ryan Dehoff, for agreeing to host me at ORNL and provide unique mentorship. His guidance enabled me to obtain hands on experience in all aspects of metallurgy and metal additive manufacturing from metallography to microscopy to machine operation. His support along the way has been intellectually challenging and helped me to grow as a researcher.

I would like to thank a large number of colleagues that have supported me along the way: Fred List for his mentorship, hardware support, and scientific approach, Frank Medina for his mentorship, processing knowledge, and collaboration, Larry Lowe for his support of hardware and sample production, Suresh Babu for his mentorship and metallurgy

knowledge, Michael Goin for his development of software tools at ORNL, Michael Pearce for his development of software tools at ORNL, Michael Kirka for his knowledge of superalloy metallurgy and collaboration, Kinga Unocic for her knowledge of microscopy and

collaboration, Don Erdman for his support of mechanical testing and collaboration, Tom Geer for teaching me metallography, Eddie Schwalbach for his collaboration, Bill Peter for his support of powder metallurgy and processing, Grant Helmreich for his collaboration, Paul Menchover for his microscopy support, Tom Muth for his metallurgy expertise, Dane Wilson for his assistance in running heat treatments, Donovan Leonard for his microscopy support, Lindsay Kolbus for her collaboration, Tapasvi Lolla for his modeling support and

(4)

iv

collaboration, Ralph Dinwiddie for his work on thermal imaging, Chad Parish for his microscopy support, Shawn Reeves for her microscopy support, Chad Duty for his support as my ORNL group leader, Craig Blue for supporting metal AM efforts at the MDF, and the rest of the great team at ORNL and TAMU that has supported me along the way. I would further like to thank my committee for their time and efforts considering my proposal, dissertation, and defense.

This research was supported by fellowship funding received from the U.S. Department of Energy, Office of Nuclear Energy, Nuclear Energy University Programs. I can truly say that this fellowship changed my life. I am thankful for the opportunities that it provided.

Portions of this research were also sponsored by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, under contract DE-AC05-00OR22725 with UT-Battelle, LLC. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes.

(5)

v

NOMENCLATURE

XZ Plane a cross-section of material taken parallel to the build direction with the build direction (z-axis) oriented towards the top of the image unless otherwise noted

XY Plane a cross-section of material taken perpendicular to the build direction

AM additive manufacturing

EBM electron beam melting

LM laser melting (sometimes used as SLM for selective laser melting)

DED direct energy deposition

IN718 Inconel 718, N07718, Alloy 718 IN625 Inconel 625, N06625, Alloy 625

YS yield strength

UTS ultimate tensile stress

FCC face-centered cubic

BCT body-centered tetragonal

GB grain boundary

𝛿, 𝛿-phase delta-phase, orthorhombic Ni3Nb phase in IN718 𝛿-needle needle-like variant of 𝛿-phase

𝛾′′ _{gamma-double-prime, BCT Ni3Nb phase in IN718} 𝛾′ _{gamma-prime, FCC Ni3(Al, Ti) phase in IN718}

(6)

vi 𝛾, 𝛾-matrix FCC nickel matrix of IN718

DS directionally solidified

SX single crystal

NR not reported (used in tables to denote unreported information) LOM light optical microscope

SEM scanning electron microscope EBSD electron backscatter diffraction

XRD x-ray diffraction

(7)

vii

LIST OF FIGURES

Page

Figure 1. LM system schematic. ... 9

Figure 2. EBM system schematic. ... 9

Figure 3. Electron beam, wire-fed DED system. ... 11

Figure 4. Laser, powder-fed DED system (LENS). ... 11

Figure 5. Process Design and Phenomena Map for metal AM. ... 14

Figure 6. Sketch of the contributions to surface finish by (a) layer roughness and (b) actual surface roughness showing satellites of small powder particles incorporated onto the surface. ... 16

Figure 7. Comparison of powder quality before use: (a) SEM 250x of GA, (b) SEM 500x of GA, (c) LOM of GA, (d) SEM 200x of RA, (e) SEM 500x of RA, (f) LOM of RA, (g) SEM 200x of PREP, (h) SEM 500x of PREP, (i) LOM of PREP. ... 19

Figure 8. An event of “smoking” caused by electrostatic repulsion: (a) distributed powder bed, (b) applied beam, and (c) “smoking” or a cloud of charged powder particles. ... 23

Figure 9. Light optical microscopy showing comparison of process induced, lack of fusion porosity to entrapped, gas porosity transferred from the powder feedstock. ... 24

Figure 10. Scan strategies used to determine heat source path in metal AM as seen in the X-Y plane (perpendicular to the build direction): (a) unidirectional or concurrent fill, (b) bi-directional, snaking, or countercurrent fill, (c) island scanning, (d) spot melting, (e) spot melting contours with snaking fill, and (f) line melting contours with snaking fill. ... 27

(13)

xiii

Figure 11. Layer delamination and cracking can be a problem in SLM (shown for M2 tool steel). ... 28 Figure 12. For EBM printed Ti-6Al-4V parts it can be seen that (Left) a NIST test artifact

designed to test AM capabilities has overhangs printed in the side of the part. These are intended to test the ability of various machines to print overhangs. Minor swelling can be seen both above the overhang and near a hole on the left side on the top surface. (Right) A complex robotic part shows a slightly deformed overhang, with sintered powder and support

stubs left underneath. ... 30 Figure 13. Melt ball formation and Delamination in EBM stainless steel. ... 31 Figure 14. The effect of substrate warping can lead to lack-of-fusion or delamination. ... 32 Figure 15. Work using neutron diffraction to measure residual stress in (a-c) LM and

(d-f) EBM IN718 shows consistently lower residual stress in EBM samples

than in LM samples. ... 35 Figure 16. The relationship between speed and current is set for EBM using the “speed

function” in an attempt to define successful processing space for the EBM process. ... 39 Figure 17. Process map for stainless steel EBM demonstrates importance in the

relationship between applied power and beam speed. ... 39 Figure 18. Relationship between effective power and speed in determining the

weldability of Inconel 718. ... 40 Figure 19. EBM processing window for Inconel 718 processing overlaid on G vs. R data. ... 42 Figure 20. Thermal simulation of a point during powder-fed DED showing cyclic heating

cycles. ... 43 Figure 21. EBM process thermal history, as measured by the machine-standard,

(14)

xiv

Figure 22. Stress relief through vacuum annealing can nearly eliminate residual stress present in as-fabricated material throughout the thickness of parts

(shown here for LM iron). ... 47

Figure 23. Recrystallization of SLM IN718 during stress relief produces an inhomogeneous grain structure. ... 48

Figure 24. Thin wall EBM fracture surface of Inconel 718 from post-HIP sample with notable change in surface oxidation and oxidation of an open pore caused by lack-of-fusion near the edge. ... 50

Figure 25. Post-HIP Ti-6Al-4V brackets, before (Top) and after (Bottom) machining. ... 52

Figure 26. Airfoil repair using a hybrid DED+CNC method. ... 53

Figure 27. Grain structure in DED material is highly influenced by scan strategy. Shown is Inconel 718, produced using (a) unidirectional, (b), bi-directional, and (c) bi-directional, high power scanning. ... 55

Figure 28. Effect of powder and edges on grain growth in EBM Ti-6Al-4V. ... 56

Figure 29. Local control of grain orientation in EBM of IN718. ... 58

Figure 30. Fatigue test results of HIP and stress relieved DED material. ... 65

Figure 31. Joint of a robotic arm that embeds hydraulic lines, eliminating external lines for hydraulic fluid and wiring. ... 70

Figure 32. Comparative analysis of additive and subtractive manufacturing. ... 72

Figure 33. Failed build due to selective powder fetching in EBM. Hardware/software advances are needed to eliminate such problems. ... 74

Figure 34. Investment casting was used to reduce the number of parts in a diffuser case from 38 to one... 82

(15)

xv

Figure 36. Dark Field TEM image and diffraction pattern showing 𝜸′′ precipitates in

IN718. ... 86

Figure 37. Formation of 𝜹-phase in IN718 by (a) cellular precipitation at <700°C, (b) cellular precipitation and transformation from 𝜸′′ at 750-800°C, and (c) direct precipitation from the 𝜸-matrix at 960°C. ... 87

Figure 38. Sketch of the typical precipitation order of 𝜸′′ and 𝜹-phase in IN718. ... 89

Figure 39. Lattice misfit and radius of 𝜸′ precipitates determine morphology (shown for a 𝜸′-strengthened alloy). ... 91

Figure 40. Formation of 𝜹-phase, 𝜶-Cr, 𝝈-phase, and Laves phase in a sample aged for 25,000hr at 676°C. ... 92

Figure 41. Solidification path for IN718, as determined in work on non-equilibrium, welding solidification. ... 94

Figure 42. Solidification diagram based on welding research. This diagram does not allow for suppression of Laves formation at higher weight percentages of niobium. ... 95

Figure 43. Relationship of primary carbide size to cooling rate for IN718. ... 96

Figure 44. Solidification map for IN718. ... 97

Figure 45. Nucleation sites shown to occur near carbides in deformed IN718. ... 99

Figure 46. Hardness of IN718 as a function of temperature. Note that peak aged material is 466HV. ... 101

Figure 47. Mechanisms for non-uniform precipitation in IN718. ... 102

Figure 48. Relationship of machinability, grain size, and hardness. Larger grains are less machinable, due to increases in work hardening from the machining process. ... 105

(16)

xvi

Figure 50. Relationship of SDAS to Yield Strength in IN718. ... 109

Figure 51. Smooth fatigue data for IN718 and derivative (PWA1472). ... 111

Figure 52. (Left) An Arcam A2 model showing an open build chamber and (Right) an internal view of the build chamber showing various components. ... 115

Figure 53. Failed F125 tungsten filament, used in an Arcam S12, showing a spot of material degradation on the filament tip. ... 117

Figure 54. (Top) Movement of focal point of e-beam using oscillatory motion, and (Bottom) the larger area over which fusion takes place within beam interaction area. (US 7454262 B2) ... 123

Figure 55. The resulting heat profile from the use of oscillatory motion or “interference”, where 𝜶 is without interference and 𝜷 is with interference. D is the diameter of the focal point. (US 7454262 B2) ... 123

Figure 56. Originally patented configuration (US 7537722 B2). The PDS as described was never implemented on a commercial system. ... 125

Figure 57. Currently implemented powder distribution system (PDS) in Arcam EBM systems. (US 7871551 B2) ... 126

Figure 58. (Top) Description of various scan strategies and (Bottom) description of melt area sub-division into small units. ... 128

Figure 59. IR camera setup for temperature monitoring. (US 20120100031 A1) ... 129

Figure 60. The change in temperature during the stages of the EBM process. ... 131

Figure 61. Four overlapping circles of radius, R, and spacing, d... 136

Figure 62. Two overlapping Gaussian distributions for modeling melt pool overlap in the X-Z direction. ... 138

Figure 63. The steps of melting; (a) contour melting using spot melts, (b) standard linear bulk melting, (c) alternative spot bulk melting, and (d) completed melt of a part slice for example geometry. ... 140

(17)

xvii

Figure 64. Melt time optimization for slices of a (a) square, (c) triangle, and (e) complex build. The standard linear melt is rotated through angles of 360 degrees

to show variation for the (b) square, (d) triangle, and (f) complex build. ... 141

Figure 65. Build Assembler software is used to create .ABF machine code to run the Arcam systems. The bottom slice of the NIST Test Artifact is shown. ... 143

Figure 66. The “Build” tab in EBM Control is used to load geometry, select start plate parameters, and save parameter inputs as a Project. ... 145

Figure 67. The “Process” tab in EBM control allows the user to load and select material sintering and processing parameters. ... 146

Figure 68. Parameters for the Preheat step. ... 149

Figure 69. Parameters for the Preheat 1 (Preheat I) step. ... 150

Figure 70. The melt.hatch.square menu is used to understand the values used for beam speed and beam current. ... 153

Figure 71. NIST Test Artifact with relevant dimensions. ... 154

Figure 72. Measured values of beam current for a representative layer. ... 155

Figure 73. Variation of melt current with line scan length in an example build. ... 156

Figure 74. Melt current used during processing of the first layer of the NIST Test Artifact example. ... 163

Figure 75. Beam speed used during processing of the first layer of the NIST Test Artifact example. ... 164

Figure 76. Applied energy density during processing of the first layer of the NIST Test Artifact example. ... 165

Figure 77. Deposited NIST Test Artifact in IN625 from example. Swelling is notable in the corner regions, which corresponds to areas of increased energy application. ... 166

(18)

xviii

Figure 78. The amount of energy is over applied when using (Top) the standard, non-linear Speed Function and (Bottom) is uniform through corners with a

linear Speed Function. ... 167

Figure 79. Over-sintering of powder in EBM IN718. ... 169

Figure 80. Average layer time as calculated by the ORNL code from log file data. ... 172

Figure 81. The “X-Y Tensile” build geometry. ... 175

Figure 82. The “Verification” build geometry (Left) with box and (Right) without. ... 175

Figure 83. Comparison of the layer time between EBM Control 3.2 (Build 3, 2014_01_15) and 4.1 (Build 4, 2014_02_20) shows increased variance in layer time with version 4.1. ... 177

Figure 84. Beam current profiles starting with the first layer of each build for (Top) Build 3, (Center) Build 4, and (Bottom) Build 7. ... 179

Figure 85. Beam current profiles for the first layer of (Top) Build 4 and (Bottom) Build 8. ... 184

Figure 86. Temperature profiles comparing standard build processes with comparable geometries (Verification Build) for IN718 (Build 6) and Ti-6Al-4V (Build 10). 186 Figure 87. (Left) SEM showing clusters of precipitate phase in interdendritic regions that (Bottom Right) are characteristic of the material as observed in a larger scale SEM image. (Top Right) This microstructure occurs near of the build (XZ shown using LOM). ... 192

Figure 88. LOM near the top surface of EBM IN718 shows two distinct region transitions: 5-10um from the surface and 30-35um from the surface. ... 193

Figure 89. Higher resolution SEM of precipitate clusters near top surface in the XY plane (perpendicular to build direction). ... 194

Figure 90. SEM of un-etched EBM IN718 in the XZ plane (parallel to build direction) showing the difference between spherical, gas pores and shrinkage porosity from solidification. ... 194

(19)

xix

Figure 91. Solidification shrinkage pores are space ~250um apart and were found to correspond to the angle of beam scanning. LOM near the top surface of a PREP, 50um, slow cool build in the XY plane. ... 196 Figure 92. Characteristic LOM near an edge in the XY plane near the top surface of a

PREP, 50um, slow cool build in the XY plane. ... 196 Figure 93. Typical grain orientation structure of IN718 using EBSD for (Left) the XY

plane and (Right) the XZ plane for a sample prepared from the Verification Build geometry (taken from the bottom of a verification build cylinder as reported in §V.2.2). ... 197 Figure 94. SEM showing precipitate networks and GBs in the XY plane. ... 199 Figure 95. Phase banding of 𝜹-needles in the XZ plane in the middle of the 50um, PREP,

fast cool, vertical tensile sample (Build 6A from §VIII.4). ... 200 Figure 96. (Left) SEM of typical “under-aged” structure with columns of carbides and

absence of 𝜹-needles and (Right) SEM of typical “over-aged” structure showing significant quantities of intragranular 𝜹-needles. (Left) From 50um, PREP, slow cool, XY Tensile build (Build 13A from §VIII.4) and (Right) from 50um, PREP, fast cool, vertical tensile samples (Build 6A

from §VIII.4). ... 202 Figure 97. Thermal history of builds showing (Top) over-aged and (Bottom)

under-aged microstructure. ... 202 Figure 98. Axial variation of 𝜹-phase showing (a) LOM and (c) SEM of etched

microstructure in the XZ plane near the top of a part (but not the top of the build) and (b) LOM and (d) SEM of the XZ plane near the bottom of

the build. ... 203 Figure 99. As-fabricated SEM of XZ planes in (a) top, (b) middle, and (c) bottom of the

50um, PREP, fast cool, vertical tensile sample (Build 6A from §VIII.4). The (d) XY plane near the bottom shows the 𝜹-needle structure more clearly. ... 205 Figure 100. Analysis of as-fabricated phase structure of overaged EBM IN718 using (a)

BF STEM, (b) high angle angular DF image (Z-contrast image), (c) higher

(20)

xx

Figure 101. Cool down from process temperature plotted on a CCT diagram. ... 211 Figure 102. Thermal histories as measured from start plate thermocouple for various

cool down techniques. The standard HT is graphed for comparison to the ISHT. ... 214 Figure 103. Builds showing (Left) a control build and (Right) the result of the in situ

heat treatment. Due to processing conditions, significant over-sintering of powder occurred during in situ processing. ... 215 Figure 104. SEM of precipitates in the matrix for (a) fast cool, (b) slow cool, and (c) in

situ heat treatment. ... 216 Figure 105. EBSD showing grain structure for a control case in the (a) XY plane and (b)

XZ plane, and for in situ heat treatment in the (c) XY plane and (d) XZ

plane. ... 219 Figure 106. SEM of etched (a) control sample (Build 16A from §VIII.4) shows lines of

shrinkage porosity aligned in z-direction and (b) in situ sample shows

aligned cracks. ... 221 Figure 107. Mechanical testing results comparing the ISHT to a control case. Two

tensile tests were completed for each case. ... 223 Figure 108. LOM of fracture surfaces for the (Left) control sample (Build 16A from

§VIII.4) and (Right) ISHT sample. ... 223 Figure 109. The hardness profile in the z-direction confirms uniform axial aging of

material; the in situ case is at or above the hardness of peak aged

material. ... 224 Figure 110. The hardness profile in the x-direction confirms uniform aging. ... 224 Figure 111. HIP cycle temperature and pressure for settings including start-up and cool

down of 980°C, 100MPa for 2hr (other HIP cycles were tested as well). ... 227 Figure 112. As-fabricated microstructure showing (a) intragranular 𝜹-needles was

processed at 1200°C, 100MPa for 2hr and (b) microstructure with no intragranular 𝜹-needles was processed with a HIP cycle of 1120°C,

(21)

xxi

100MPa for 2hr. The results are (c) large grain growth and the disruption of the oriented, columnar structure and (d) limited coarsening and no disruption of the grain structure. Images were taken of (a) SEM of the XZ plane, (b) SEM of the XZ plane, (c) EBSD of the XY plane, and (d) EBSD of

the XY plane. ... 229 Figure 113. Etched LOM showing persistence of shrinkage pores in the (a) XY, (b) XZ,

and (c) XZ planes after HIP at (a) 100MPa, 1120°C, 120min, (b) 100MPa,

800°C, 120min, and (c) 200MPa, 800°C, 120min. ... 230 Figure 114. SEM of the XZ planes of as-fabricated microstructures (1A) without

intragranular 𝜹-needles and (8A) with needles show variation in the resulting phases present given the same solution treatment and aging of 980°C for 1hr, 720°C for 8hr, and 620°C for 10hr that shows (1B) no precipitation of 𝜹-needles and (8B) no dissolution and possible growth of existing 𝜹-needles. ... 232 Figure 115. LOM of the XZ plane shows (8M) complete 𝜹-needle solution and minor

grain growth or RX, (8N) significant grain growth, (8P) incomplete solution of 𝜹-needles, and (8Q) significant grain growth. ... 234 Figure 116. EBSD results of the ST matrix show significant grain growth in the 1200°C,

4hr case but do not show significant difference in grain orientation in the 1120°C, 2hr case. The As-Fabricated sample is from the bottom of Build

8A from §VIII.4... 235 Figure 117. SEM image of ST IN718 at 1020°C for 1hr (8P). Potential RX nucleation sites

or early stage growing grains are identified with arrows. ... 236 Figure 118. Area fraction of various grain diameter bin sizes for (8A) as-fabricated and

(8M) ST at 1120°C for 2hr material. ... 237 Figure 119. Grain boundaries (yellow), primary twin boundaries (red), and secondary

twin boundaries (blue) are identified in the XY plane of sample 8N. ... 239 Figure 120. OIM analysis of EBSD data from the XY plane of sample 8N was used to

map (a) unique grains including annealing twins and (b) unique grain

(22)

xxii

Figure 121. SEM of the XZ plane of (Left) ST at 1120°C for 2hr and (Right) ST at 1200°C for 4hr shows that carbides remain in a columnar structure in the build direction in the lower temperature/time case but exist in a more isotropic distribution in the higher temperature/time case with significant grain

growth. ... 242 Figure 122. SEM of the XY plane of (Left) ST at 1120°C for 2hr and (Right) ST at 1200°C

for 4hr shows that significant numbers of carbides have fallen out of the higher temperature ST sample during metallography, which complicates analysis. ... 242 Figure 123. SEM of the XY plane of 8Q shows persistence of solidification porosity after

ST (aligned in the center of the image) and the smaller carbide “holes” where large, coarsened carbides have fallen out of the sample during

metallography. ... 243 Figure 124. By grouping carbides into a histogram, the increase in area fraction of large

carbides (>1um) in the high temperature ST (1200°C for 4hr) becomes apparent. Carbides are assumed to be circular and are grouped according to a calculated diameter. ... 244 Figure 125. Possible mechanisms for RX during heat treatment of (a) overaged IN718

with columnar grain structure include (b-d) the Carbide PSN mechanism

and the (e-g) Needle PSN mechanism. ... 246 Figure 126. TEM showing dislocations near 𝜹-phase in previous work that noted PSN

assisted dynamic RX in IN718. ... 248 Figure 127. Flow diagram showing the relationship of processing, microstructure, and

impact on the final material for EBM. ... 251 Figure 128. Comparison of powder quality before use: (a) SEM 250x of GA, (b) SEM

500x of GA, (c) LOM of GA, (d) SEM 200x of RA, (e) SEM 500x of RA, (f) LOM of RA, (g) SEM 200x of PREP, (h) SEM 500x of PREP,

(i) LOM of PREP... 253 Figure 129. Threshold analysis shows greatly reduced porosity from PREP samples

(23)

xxiii

Figure 130. (a) New PREP powder at 200X, (b) used PREP powder after ~420 hours at 200X, (c) New PREP powder at 500X, and (d) used PREP powder after

~420 hours at 500X. [20] ... 255 Figure 131. The equipment used at ORNL to perform tensile testing. ... 256 Figure 132. Representative stress-strain tests show the age hardening effect of cool

down. ... 260 Figure 133. Etched SEM microstructure in the XZ plane of 16A shows aligned carbides

and shrinkage porosity in the build direction. ... 262 Figure 134. Thermal histories for Builds 7A and 10A shows the speed of the 75um layer

thickness with parameters optimized for build speed (Build 10A). ... 264 Figure 135. Tensile results of horizontal samples from 15A tested at various z-heights

show little variation in UTS, YS, or elastic modulus with sample height. ... 267 Figure 136. Tensile results of horizontal samples from 20A tested at various z-heights

show improvement in UTS, YS, and elongation for material closer to the

top of the part. ... 268 Figure 137. Comparison of the thermal histories for Verification Builds 15A and 20A

shows that 20A had lower hold temperature and longer hold time than

did 15A... 269 Figure 138. Stress vs. strain response for representative tensile samples for the PREP

slow cool (13A) and GA slow cool (Build #2) cases shows limited variation in YS, UTS, and elongation. ... 270 Figure 139. The thermal histories of the PREP slow cool (13A) and GA slow cool (Build

#2) cases show an increased hold time for 13A. The hold temperature for both builds remains ~1000°C. ... 271 Figure 140. Representative stress-strain response shows STA results in higher UTS/YS

and lower elongation for all cases tested. Each plot is a direct comparison between the as-fabricated and STA material from the same batch. ... 273

(24)

xxiv

Figure 141. (Top) Representative tensile tests across samples 1, 2, 7, 8, 14 and

(Bottom) representative STA tensile results of the same samples. ... 274 Figure 142. Results of YS and UTS show that as-fabricated EBM material nearly meets

the ASTM standard for stress relieved IN718 from powder bed fusion and that, after heat treatment, EBM STA material exceeds the standard for

UTS/YS. ... 275 Figure 143. As-fabricated EBM IN718 nearly meets the ASTM standard for elongation

for stress relieved powder bed fusion material and exceeds the standard for elongation after STA. ... 276 Figure 144. (Left) Turbine blade made of IN718 using EBM at ORNL with (Right)

internal features exposed by cutting through the sample cross-section. ... 279 Figure 145. Analysis of process parameters shows that (Left) increases in local applied

energy can lead to (Right) swelling in EBM parts (IN625 pictured). ... 281 Figure 146. The formation of intragranular 𝜹-phase needles typically happens in

samples where build temperature falls below the temperature for direct 𝜹-phase precipitation from the matrix (~960°C). ... 282 Figure 147. Cool down procedure was shown to effect significant changes in hardness. .. 283 Figure 148. An (b, e) optimized solution treatment can be used to dissolve 𝜹-needles

without incurring significant grain growth as seen at (c,f) higher

temperatures. ... 285 Figure 149. EBM IN718 nearly meets ASTM F3055 standards by the F42 committee for

stress relieved material directly from the machine. Heat treated EBM

(25)

xxv

LIST OF TABLES

Page

Table 1. Original patents for the various classifications of metal AM. ... 6 Table 2. Typcial layer thicknesses and minimum feature sizes of PBF and DED

processes. ... 15 Table 3. Common post-processing procedures for Ti-6Al-4V and Inconel 718. ... 46 Table 4. Compilation of reported tensile results for Ti-6Al-4V and Inconel 718. ... 62 Table 5. Comparison of defects and features across platforms. ... 67 Table 6. Reported deposition rates for various technologies ... 69 Table 7. ASTM Standard on IN718 chemistry. ... 83 Table 8. Standard vs. Historical Terminology for IN718. ... 84 Table 9. Possible phases in IN718 with reference lattice parameters and solvus

temperatures. ... 85 Table 10. Measured composition of IN718 phases. ... 90 Table 11. Tensile properties of cast and wrought IN718. ... 109 Table 12. Stress rupture data for IN718. ... 111 Table 13. Hardware specifications for various Arcam models. ... 116 Table 14. Patents held on Electron Beam Melting (EBM) by Arcam AB. ... 120 Table 15. Select Process Parameters for the NIST build example. ... 159

(26)

xxvi

Table 16. The measured process data compared to the nominal or calculated

parameter inputs. ... 160 Table 17. Summary of analyzed builds. ... 173 Table 18. Summary layer time, process time, and deposition rate. ... 174 Table 19. Breakdown of layer time for the first layer of select builds. Data given in

both time and percentage of total layer time. ... 180 Table 20. Changes in process parameters between Builds 3, 4, and 7. ... 181 Table 21. Differences in process parameters between Build 4 and Build 8. ... 182 Table 22. Process current used to achieve in situ heat treatment. ... 213 Table 23. Hardness values considered for all x- and z-positions. ... 222 Table 24. Results and nomenclature of ST test matrix (Build labels correspond to heat

treatments of the starting material from Build 8A from §VIII.4). ... 233 Table 25. Measured hardness of as-fabricated and solution treated material. ... 240 Table 26. Tensile results for as-fabricated EBM IN718. ... 258 Table 27. Tensile test results for standard heat treatment of EBM IN718. ... 272

(27)

1

CHAPTER I

INTRODUCTION

I.1 Additive Manufacturing

Additive Manufacturing (AM), widely known as 3D printing, is a method of manufacturing that forms parts from powder, wire, or sheets in a process that proceeds layer-by-layer. Many techniques (using many different names) have been developed to accomplish this via melting or solid-state joining. Direct energy deposition (DED) processes feed powder of wire feedstock into a melt pool to produce parts. Electron Beam Melting (EBM) and Laser Melting (LM) are both powder bed fusion (PBF) processes, that selectively melt distribute layers of metal powder to produce a part from the bottom up. PBF processes have the advantage of being able to produce more complex shapes and features (e.g. overhangs) than DED processes. All metal AM processes have advantages over casting and subtractive machining in that unique geometries can be produced, there is a low cost of retooling, and small production quantities can be more economical.

This dissertation focuses on EBM because there has been recent interest from industrial part producers in utilizing the technology. The EBM technology has faster deposition rates (§II.7) and lower amounts of residual stress (§II.3.10) than LM. The energy density of the heat source (§II.7) also enables complete melting of a wide range of alloys. Additionally, the use of a vacuum during processing reduces many issues with oxidation during processing (§II.3.2). Finally, the ability to engineer grain orientations has demonstrated significant potential for the technology to improve materials design and engineering (§II.6.1.3).

(28)

2

I.2 Importance for Advanced Nuclear Energy Concepts

Inconel 718 is noted for maintaining its mechanical strength at elevated operating temperatures and is used in a variety of applications (§III.1). Of specific interest (as this dissertation was the focus of a Nuclear Engineering program), is the application of Inconel 718 into power production systems. Various in-core components for nuclear reactors and components on the secondary side (tubes, heat exchangers, turbine blades) either use or have the potential to use this material. By enabling studying and improving a method of AM for IN718, this dissertation asserts that this EBM technology presents an opportunity for designers of nuclear components to produce unique designs to improve efficiency or safety of reactor systems.

I.3 Overview of Work Presented Herein

The key findings of each chapter in this dissertation can be summarized as:

 Chapter II – Reviews the whole of metal Additive Manufacturing, offering insights into the differences between various technologies and background on processing science

 Chapter III – Reviews Inconel 718, providing a concise background on the material that is the focus of this paper

 Chapter IV – Focuses on the Arcam Electron Beam Melting (EBM) hardware and provides background on the experiments run for this dissertation

 Chapter V – Explains EBM processing parameters and the impact that changes to those parameters had on the measured process time and process temperature

 Chapter VI – Characterizes the as-fabricated microstructure of EBM IN718 and presents a novel method for control phase formations in situ

(29)

3

 Chapter VII – Explores ex situ post-processing options for improving mechanical properties and rationalizes grain growth phenomena that were observed

 Chapter VIII – Reports the differences in as-fabricated tensile properties and the use of solution treatment and aging to achieve uniform tensile properties

(30)

4

CHAPTER II

REVIEW OF METAL ADDITIVE MANUFACTURING

This chapter has been adapted from “The Metallurgy and Processing Science of Metal Additive Manufacturing” by Sames et al., an invited review to the International Materials Reviews. [1]

In this chapter, AM techniques for producing metal parts are explored, with a focus on the science of metal AM: processing defects, heat transfer, solidification, solid-state

precipitation, mechanical properties, and post-processing metallurgy. The various metal AM techniques are compared, with analysis of the strengths and limitations of each. Few alloys have been developed for commercial production using AM, but recent efforts are presented as a path for the ongoing development of new materials for AM processes.

II.1 Introduction & History

Additive Manufacturing (AM), or 3D printing, has grown and changed tremendously in the past 30 years since researchers in Austin, TX started development of what is arguably the first machine in the lineage of metal AM: a laser used to selectively melt layers of polymer and, later, metal. [2] The development of metal AM techniques has made great progress since then, but faces unique processing and materials development issues. Understanding the various processes used to make metal AM parts, and the issues associated with them, is critical to improving the capabilities of the hardware and the materials that are produced.

The first experiments with metal AM grew out of efforts originally targeted at forming polymer powder into 3D parts. [3-6] This research focused on powder bed laser sintering, which was patented and copyrighted as Selective Laser Sintering (SLS). One of the earliest incarnations of SLS, “Betsy”, integrated the first automated powder distribution system. SLS

(31)

5

by similar techniques is still used today to produce metal parts (the term “sintering” is now used loosely, as many processes use complete melting) under license by EOS GmbH. Shortly after SLS was patented, a group of researchers at MIT patented a process called “three-dimensional printing”, which used inkjet printing to deposit binder. The use of “3D

printing” has grown to describe all forms of AM, while the MIT method has become known as Binder Jetting (BJ). BJ can be used to create metal parts, in addition to other materials. Another class of printers known as Direct Energy Deposition (DED) deposit feedstock

directly into a molten pool, as opposed to selective melting of a powder bed. Some of these machines use a wire feedstock and are very similar to welding processes. In 1995, Sandia National Laboratories developed a different approach to powder feedstock in DED with a laser heat source. This technology was first commercialized and trademarked as Laser Engineered Net Shaping (LENS), a sub-set of DED. The last major category of metal AM, sheet lamination (SL), welds together sheets of feedstock to form parts. A process that uses ultrasonic welding and computer numerical control (CNC) milling to accomplish this was originally developed and patented by Dawn White of Solidica in 1999. In 2000, research in Sweden led to the patent of another powder bed technique: Electron Beam Melting (EBM). This process was later licensed and developed by Arcam AB. This metal AM history is more concisely presented as a timeline), with significant patents highlighted in Table 1.

II.1.1 Timeline: “30 Years of 3D Metal Printing”

 1984 – Deckard & Beaman begin work on technology to build 3D parts out of powder, using a 100W YAG laser heat source [3]

 1986 - Deckard & Beaman start “Betsy”, continue research into “SLS” [3]

 1986 – SLS patented by Deckard at University of Texas [7]

 1989 - Original “3D Printing”, or inkjet binder deposition, patented by Sachs & Cima at MIT [8]

(32)

6

 1995 – Sandia National Laboratories begins LENS development [10]

 1997 – EOS licenses SLS rights from 3D Systems and focuses on powder bed technology [9]

 1999 – Ultrasonic consolidation patented by Dawn White of Solidica [11]

 2000 – EBM patented by Andersson & Larson [12]

 2001 – 3D Systems acquires rights to SLS technology through acquisition of company holding original patents by Deckard [3]

 2002 – Arcam launches first commercial machine, the S12 [13]

 2007 – CE-certification of hip implant manufactured by EBM [13]

 2012 – 3D Systems acquires Z-Corp, holder of original patent on inkjet binder process [3]

Table 1. Original patents for the various classifications of metal AM.

Process Patent Number Patent

Priority Date

Inventor Original Associated Corporation SLS [7] WO1988002677 A2 1986 Carl Deckard University of

Texas “3D Printing” [8]

(binder jetting)

US5204055 A 1989 Cima et al. MIT

EBM [12] US 7537722 B2 2000 Lars-Erik Andersson & Morgan Larsson

Arcam AB

LENS [14] US 6046426 A 1996 Jeantette et al. Sandia Corporation

UAM [11] US6519500 1999 Dawn White Solidica

Since the invention of the various metal AM processes, R&D and industry efforts have developed some niche applications. Part repairs, biomedical implants, aerospace

structures, and high temperature components highlight some of the current production use of the technologies. Metal AM has received increasing attention for direct production of

(33)

7

end-use parts, even being highlighted by U.S. President Barack Obama in a 2014 speech on manufacturing. [15] Despite the recent attention, some big questions remain: What are the current limitations of the technology? Can those limits be overcome?

II.2 Classification of Technologies

A diverse set of processes can be used to form feedstock (powder, sheets, or wire) into 3D objects. All metal AM processes must bond together the feedstock into a dense part. The metal must be melted at some point in the process to achieve this. In order to discuss distinct classes of machines, the ASTM F42 Committee on Additive Manufacturing has issued a standard on process terminology. [16] Of the seven F42 standard categories, the following four pertain to metal AM:

 powder bed fusion (PBF) o laser melting (LM)

o electron beam melting (EBM)

 direct energy deposition (DED) o laser vs. e-beam o wire-fed vs. powder-fed  binder jetting (BJ) o infiltration o consolidation  sheet lamination (SL)

o ultrasonic additive manufacturing (UAM)

The other three categories specified in the standard do not currently apply to metal technologies: material extrusion, material jetting, and vat photopolymerization. There are unique uses, strengths, and challenges for each process. Each category for metal AM is

(34)

8

explored, but more depth is given to DED and PBF due to the larger volume of recent work published on those processes.

II.2.1 Powder Bed Fusion

Powder bed fusion (PBF) includes all processes where focused energy (electron beam or laser beam) is used to selectively melt or sinter a layer of a powder bed. For metals, melting is typically used instead of sintering. The use of laser sintering has been previously

reviewed, [17] but much progress has been made since this work to include the use of full melting. Re-melting of previous layers during the melting of the current layer allows for adherence of the current layer to the rest of the part. A schematic of a PBF laser melting (LM) machine is shown in Figure 1. A schematic of PBF electron beam melting (EBM) is shown in Figure 2. Although both systems use the same powder bed principle for layer-wise selective melting, there are significant differences in the hardware setup. The EBM system is essentially a high-powered scanning electron microscope (SEM), which requires a

filament, magnetic coils to collimate and deflect the position of the beam, and an electron beam column. LM typically has a system of lenses and a scanning mirror or galvanometer to maneuver the position of the beam. Powder distribution is handled differently as well; LM systems use a dispersing piston and roller, while EBM systems use powder hoppers and a rake. Both EBM and LM processes require certain steps: machine setup, operation, powder recovery, and substrate removal.

(35)

9

Figure 1. LM system schematic. [18]

(36)

10

To setup a PBF machine, the build substrate must be positioned. The build substrate, or “start plate”, is used to give mechanical and thermal support to the build material. LM processes bolt or clamp down the substrate, whereas the EBM process typically sinters powder surrounding the plate to provide stability. When successive layers of powder are distributed (rolled or raked out), existing layers of the build must not move. The substrate helps eliminate swelling and other process defects in PBF from the first layers, as building overhangs on top of loose powder can cause localized temperature fluctuations. Finally, powder containers must be loaded and a number of sensors checked and adjusted.

The operation of a PBF machine is governed by the details of the scan strategy (§II.3.6) and processing parameters (§II.4.3), which will be discussed later in more detail. After the build is complete, excess powder must be removed from the build chamber. For EBM parts, this powder is run through a powder recovery system to remove and recover sintered powder from around the parts. For LM processes, powder surrounding the parts does not sinter as much and can be sifted directly to remove any sintered clusters. Depending on the PBF process material, the build substrate may adhere to the parts. [20] The substrate must be cut off, with abrasive saws and wire EDM being common methods. For some material combinations like Ti-6Al-4V deposit and stainless steel substrate in EBM, material properties promote poor adherence; the parts fall off the substrate after the build, or can be easily removed by applied force. Parts coming directly out of the machine are considered “as-fabricated”.

II.2.2 Direct Energy Deposition

Direct Energy Deposition (DED) encompasses all processes where focused energy generates a melt pool into which feedstock is deposited. This process can use a laser, arc, or e-beam heat source. The feedstock used can be either wire (Figure 3) or powder (Figure 4). The origins of this category can be traced to welding technology, which deposits material outside of a build environment by flowing a shield gas over the melt pool.

(37)

11

Figure 3. Electron beam, wire-fed DED system.[21]

Figure 4. Laser, powder-fed DED system (LENS). [22]

One of the most studied and commercialized forms of DED is accomplished using a laser heat source to melt a stream of powder feedstock (powder-fed). This technology sub-set

(38)

12

has its roots in research at Sandia National Laboratories and was originally patented as the LENS process.[22, 23] Other DED processes feed wire into a molten pool (wire-fed), and are essentially extensions of welding technology. [24, 25] In fact, the use of welding machines to make parts via multi-pass welding is presently being explored. [26]

Machine setup is relatively simple; machine software automatically checks most sensors. As in PBF, powder hoppers must be filled and a build substrate positioned. The substrate can be positioned in a stationary position (3-axis systems) or on rotating axes (5+ axis systems) to increase the ability of the machine to process more complex geometries. In powder-fed systems, the feed rate of the powder must be verified regularly. If flow is impeded, nozzle cleaning or other maintenance may be performed. The build chamber is enclosed to provide laser safety, but the chamber is not necessarily filled with inert gas. For non-reactive metals, a shield gas directed at the melt pool may provide enough resistance to oxidation. For more reactive metals, like titanium, the chamber is flooded with an inert gas (argon or nitrogen). A vacuum pump and purge cycles may be used to reduce oxygen content. Cyclic purging can consume a significant amount of gas, as the build chamber is much larger than those in PBF systems.

As in PBF, a finished DED part is typically attached to the build substrate. Parts are then post-processed both thermally (to reduce residual stress and improve properties) and to achieve the desired final geometry (parts produced using DED are typically near net shapes with a rough finish). Parts may be removed from the substrate using the same processes for an adhered PBF part. Excess powder from machine operation is vacuumed to clean out the machine. Depending on the operator, this powder can may be recovered or disposed. Disposal is usually a costly option, as powder costs are typically high.

(39)

13

II.3 Material Processing Issues

Although PBF and DED processes have significant differences, there are some common materials processing issues that occur in both platforms. These issues are explored, noting differences between categories of equipment where appropriate. As with traditional processing methods (casting, welding, etc.), porosity is a common concern in metal AM. Other defects (residual stress, delamination, cracking, swelling, etc.) are more unique to welding or metal AM. Scan strategy, process temperature, feedstock, build chamber atmosphere, and many other inputs determine the occurrence and quantity of defects. Understanding defects, and how they arise, can help operators improve process reliability and the quality of parts produced.

In order to understand the complex relationship between basic processing science, defects, and the end product of an AM process it is useful to consider a general process flow chart (Figure 5). This process design and phenomena map links key outcomes to process

selection decisions through fundamental phenomena. The process inputs are AM Software & Part Geometry, Scan Strategy, AM Hardware, Build Chamber Atmosphere, and Feedstock Quality. The process outputs are Mechanical Properties, Failed Builds, and Feature Size & Geometry Scaling. A box encloses the thermal and particle physics interaction steps: Applied Energy, Beam Interactions, Heat Transfer, and Process Temperature. These physics interactions, if properly modeled, should be able to describe dynamic process temperature, which is one of (if not the most) defining quantity of metal AM processing.

(40)

14

Figure 5. Process Design and Phenomena Map for metal AM.

II.3.1 Feature Size, Surface Finish, and Geometry Scaling

When printing metal parts, the minimum feature size, surface roughness, and geometrical accuracy of the part are typical concerns for equipment operators, but over-focusing on these properties is not useful for most applications because the part surface will ultimately be machined during post-processing. The minimum feature size is determined by the minimum diameter of the heat source and the size of the feedstock. This data is

(41)

15

the resolution of LM slightly better than EBM depending on parameters used. Powder-fed DED has better resolution than wire-fed DED, which can be attributed to the use of finer feedstock (powder vs. wire). The feature size of DED systems is so large that parts made with these techniques are limited to more simple geometries than PBF techniques. Smaller feature sizes smaller layer thickness currently comes at the expense of deposition rate. The deposition rates of various technologies are explored in more detail later in this paper (§II.7). Due to small feature size and the low inertia to changing the position of the beam, PBF techniques can utilize the minimum feature size to print metal mesh or foam

structures. These structures melt metal “struts”, typically the size of an individual pulse of the heat source (see Table 2). Mesh parts have been well studied and reviewed elsewhere. [27, 28]

Table 2. Typcial layer thicknesses and minimum feature sizes of PBF and DED processes.

Process Typical Layer Thickness [µm]

Minimum Feature Size or Beam Diameter [µm]

PBF – LM [29] 10-50 75-100

PBF – EBM [19] 50 100-200

DED – Powder Fed [30] 250 380

DED – Wire Fed [31] 3,000 16,000

There are two separate contributors to surface roughness as shown in Figure 6: (1) non-flat layer edges or layer roughness and (2) the actual roughness of the metal surface. The layering effect can be reduced by using smaller layer thickness values. This usually means longer melt times; the layer thickness dictates how many layers a part is divided up into. More layers translate to longer build times. The actual roughness of a material depends upon the details of the machine producing the part. DED typically has larger layer thickness, which mostly limits this technology to near net shapes (shapes produced close to the

desired part geometry, but planned for the use of machining to deliver the final geometry and details). Near net shape processing is different from traditional subtractive methods where a full block of material is machined down to a final part. PBF systems typically have

(42)

16

finer resolution and layer thickness, but are prone to satellite formation [32] due to the sintering of powder at the part edges. Finer powder means smaller satellites and less surface roughness. LM machines use finer powder and smaller layer thickness than EBM, which results in less surface roughness.

Figure 6. Sketch of the contributions to surface finish by (a) layer roughness and (b) actual surface roughness showing satellites of small powder particles incorporated onto the surface.

Geometrical accuracy can be measured by taking 3D lasers scans (or similar technique) and calculating the deviation relative to the original part file. Typical corrections are empirical modifications to scale part files in a Cartesian system. For example, an x-axis might be smaller than intended by some scaling factor. The scaling factor is then used to increase the x-axis length in the part file, before printing. This is typically accounted for during machine calibration. Post-fabrication machining is typically needed for LM, EBM, and DED parts, as even the best achievable surface finish is still not as good as a machined finish. If machining is used, the actual part tolerance, surface finish, and minimum feature size of AM parts is dictated by any machining. For this reason, work to refine surface finish using smaller powder particles and smaller layer thicknesses may just add process time and cost (the

(43)

17

smaller the layer, the more layers must be processed) without improving the quality of the final part.

II.3.2 Build Chamber Atmosphere

The atmosphere under which metal is processed strongly affects chemistry, processability, and heat transfer. Inert gas and vacuum systems are typically used, and each has unique processing concerns. Metal powders have a tendency to oxidize and collect moisture when exposed to air. At higher temperatures, this oxidation can be accelerated. For this reason, welding machines use inert shield gases. AM processes have the same need. As discussed previously, DED typically operates with a shield gas flowing over the melt surface and may operate under inert atmosphere. LM processes are typically run in an inert environment, with an atmosphere of Argon or Nitrogen filling or flowing over the build surface. The flow rate of the fill gas and the pathway of the flow have been shown to be important in porosity reduction in LM Ti-6Al-4V. [33] Small features may lead to heat concentration in LM, which can cause localized oxidation.

The EBM process uses a heated filament (usually made of tungsten) to generate electrons, which requires a vacuum-capable build chamber to operate the machine (<5x10-2 Pa chamber pressure, <5x10-4 Pa column pressure). During beam operation, a small quantity of helium is injected to reduce electrical charging of the build volume. This raises the pressure of the build chamber to ~0.3 Pa during beam operation. Operating in a near-vacuum environment leads to increased melt vaporization and unique heat transfer considerations.

II.3.3 Feedstock Quality

The quality of the feedstock that is used in the AM process is important to the quality of the final part. The quality of the powder is determined by size, shape, surface morphology,

(44)

18

composition, and amount of internal porosity. The quality of powder determines physical variables, such as flowability and apparent density. There are a variety of atomization techniques for producing metal powder, each producing distinct variations in powder quality. There are several unique quality issues related to wire feedstock for DED as well. By understanding feedstock quality, an operator can select the optimal material for processing in a given system. Further information on the standards associated with quantifying powder characteristics and the details of powder science are well described elsewhere. [34]

The quality of powder is directly related to the production technique. A variety of techniques are used: gas atomization (GA), rotary atomization (RA), plasma rotating electrode process (PREP), plasma atomization (PA), and others. Some atomization

techniques yield irregular shapes (like RA), others have a large amount of satellites (like GA), and some are highly spherical and smooth (like PREP and PA). Figure 7 shows powder surface morphology and shape, as well as cross-sections to analyze internal porosity. Porosity in the powder feedstock is common for certain production techniques, like gas-atomization (GA), that entrap inert gas during production. This entrapped gas is transferred to the part, due to rapid solidification, and results in powder-induced porosity in the

fabricated material. These pores are spherical, resulting from the vapor pressure of the entrapped gas. Higher quality powders produced via the plasma rotating electrode process (PREP) do not contain such pores and have been used to eliminate powder-induced porosity in DED and PBF systems. [20, 35, 36]

(45)

19

Figure 7. Comparison of powder quality before use: (a) SEM 250x of GA, (b) SEM 500x of GA, (c) LOM of GA, (d) SEM 200x of RA, (e) SEM 500x of RA, (f) LOM of RA, (g) SEM 200x of PREP, (h) SEM 500x of PREP, (i) LOM of PREP. [20]

Flowability (how well a powder flows) and apparent density (how well a powder packs) are important quantitative powder characteristics that are directly related to qualitative characteristics. A Hall Flow meter can be used to measure flow rate (flowability) [37] and apparent density [38] according to ASTM standards. Spherical particles improve flowability and apparent density. Smooth particle surfaces are better than surfaces with satellites or other defects. Fine particles, or “fines”, typically improve apparent density and flowability, but can become segregated from coarser particles during use.

(46)

20

The nominal particle size distribution of powder used in LM is 10-45um, in EBM is 45-106um [39], and in DED is 20-200um. [40] The main tradeoff in the selection of powder size is cost versus surface finish. Smaller particles tend to improve surface finish and are normally used with smaller layer thicknesses, but smaller powder particles typically cost more as a

feedstock (than a larger size range) due to lower yields for smaller particles in powder production. LM uses a fine distribution of powder to improve surface finish by enabling shorter layer thicknesses. EBM uses slightly thicker layers and a correspondingly large size distribution. EBM can use smaller size distributions, with no noticeable effect on chemistry, material properties, or microstructure. [41] The effect of powder flowability on

processability using various hardware is not well published, though it is understood as an important parameter by industrial producers of AM parts. PBF systems typically have a hardware specific flowability that depends on the powder distribution method used. Very fine particles size distributions that do not have a measurable flowability may still be processable in some systems. Powder-fed DED systems must consider the effect of

flowability on the ability of powder to feed into the carrier gas stream. Once in the stream, the powder flow rate has been observed to have little effect on particle speed during DED processing. [42]

Additionally, the chemical composition of the powder must remain within alloy-specific specifications. It is important to measure the elemental composition of recycled powder (wire is not recycled), as evaporative losses, contamination from powder recovery (vacuums or grit blaster used in EBM), and reaction with oxygen, nitrogen, or other gases must be considered for quality control. Depending on the feedstock material, oxidation and humidity control may be important for both wire and powder storage. A thorough survey of the many powder types used for laser processes exists [43], regarding the research available on specific powder alloys.

(47)

21

Wire feedstock for wire-fed DED processes is less defect prone than powder and is

commonly available from welding suppliers. The diameter of wire used for wire-fed DED is typically on the order of 2.4mm. [31] Better quality wire will have less variation in wire diameter, which is similar to requirements for plastic extrusions printers that use plastic wire as a feedstock. Porosity is a common welding defect, and the quality of wire is known to affect the amount of porosity in the weld deposit. [44] For reactive metal like titanium, surface adsorption and reactions with atmosphere may also cause defects. More notably, the presence of cracks or scratches on the wire surface may translate directly to porosity formation. Unlike powder production, gas porosity is not an issue in wire production. In a study of both powder and wire feedstock, it was noted that powder had porosity, whereas the wire did not. [45]

II.3.4 Beam-Powder Interactions

The interaction of the heat source with the feedstock or melt pool impacts the amount of energy utilization and can lead to ejecta and porosity. There are four basic modes of particle ejection during beam melting processes: (1) convective transport of liquid or vaporized metal out of the melt pool (or spatter ejection), (2) electrostatic repulsion of powder particles in EBM, (3) kinetic recoil of powder in DED, and (4) enhanced convection of powder in gas streams. Lasers incur intensity losses due to reflection, whereas e-beams incur backscatter losses of electrons. E-beams systems must be designed to reduce

electrical charge buildup. DED systems must also be designed to consider the effective feed rate of the feedstock, as appropriate amounts of deposit material must be delivered.

The convective transport of liquid or vapor out of the melt pool is commonly called “spatter” or “spatter ejection” and is seen in PBF, DED, and welding. This is caused by the application of a high energy beam creating localized boiling, where the energy of the ejecta must overcome surface tension forces. [46] These particles can be identified in PBF and DED

Additive Manufacturing of Inconel 718 using Electron Beam Melting: Processing, Post-Processing, & Mechanical Properties